A decade of achievement: now celebrating its 10th year, FHWA's Advanced Research program has produced a host of tangible results ranging from "smart bridges" to detectors that alert drowsy drivers.

Milton "Pete" Mills

High-risk, high-payoff research is what the Advanced Research program is all about. Among its accomplishments to date are innovative sensors for "smart bridges," new methods for measuring the development of strength in concrete, advanced traffic flow simulations using cellular automata, and the application of neural networks to detect drowsy drivers.

This year marks the 10th anniversary of the Federal Highway Administration's (FHWA) Advanced Research program. Located at Turner-Fairbank Highway Research Center, the program is under the auspices of FHWA's Office of Research, Development, and Technology. The Inter-modal Surface Transportation Efficiency Act (ISTEA) of 1991 authorized the program by mandating the implementation of "an effective high technology applied research and development program."

In response, FHWA created the Office of Advanced Research in March 1992. Since then, the exact meaning and scope of the term "advanced research" has been the subject of debate. But at least this much is clear: The Congressional intent was that this type of research was to be different from that traditionally performed by FHWA, which has generally focused on developing products that can be delivered to practitioners within a year or two.

Congress also clearly intended that the program would bring in ideas and expertise from fields outside the highway research community. Some examples were specified in the FHWA Order M 1100.1 that created the office: "advanced statistical and computational methods . . . artificial intelligence and expert systems . . . operational analysis . . . ceramics . . . robotics . . . material microstructure systems . . ." This multidisciplinary approach is similar to programs in other mission-oriented Federal agencies such as the U.S. Department of Defense, where exploratory research provides "new and improved functional capabilities" in support of the applied research programs that develop the technologies to carry out the agency's mission.

In 1998, the Transportation Equity Act for the 21st Century (TEA-21) re-authorized FHWA's Advanced Research program for an additional 6 years, and the program is under consideration for the next reauthorization legislation due in 2004.

The Transportation Research Board's (TRE) Research and Technology Coordinating Committee recently gave strong support for continuing this fundamental research activity. "One of the most important roles of FHWA is to undertake targeted high-risk advanced research deemed to be of national significance," says Dr. C. Michael Walton, chair of the TRB committee and professor at the University of Texas at Austin. "Research that is successful may directly or indirectly lead to a major advance in the knowledge base of the profession or the state of practice. Little if any advanced highway research is currently being conducted nationwide. Therefore the committee recommended that a goal of 25 percent of the research program budget be allocated to advanced research."

Funding and Staff

ISTEA specified that not less than 15 percent of the funds made available by Congress "shall be expended on long-term research projects [that] are unlikely to be completed within 10 years." The Advanced Research budget has never come close to the 15 percent level, but rather has averaged around $2 million per year. TEA-21 severely cut the research funds in fiscal year 1999, but they returned to normal levels in fiscal year 2002. In addition, partners such as the National Science Foundation and State departments of transportation have provided support amounting to more than $3 million during the 10-year period.

Over the years, the staff of the Advanced Research program has numbered between three to six. Federal employees, supplemented by in-house contractors, graduate students, and professors on sabbatical under the Intergovernmental Personnel Act program, as well as four postdoctoral fellows supplied by the National Academy of Sciences' Resident Associates Program. This reliance on transient researchers has some advantages, making it easier to bring in members of other disciplines, ranging from mathematicians to nuclear engineers, and to get up to speed more rapidly in new research areas.

In 1995, the original Office of Advanced Research was split up as part of an overall reorganization of FHWA's research and development program. Staff members were divided between the Office of Infrastructure R&D and the Office of Safety In light of the continuing Congressional mandate, the individual staff members have maintained close cooperation as a "virtual" Advanced Research team.

The 10th anniversary provides an opportunity to review what the program has accomplished in its first decade. A few examples can give an idea of its operations, though not covering all of its research activities exhaustively.

Magnetostriction for Suspension Bridge Inspection

Periodic inspection of the cables on suspension bridges is essential for safety. The current practice, however, involves unwrapping the cables and separating the individual strands using wedges. This work is time-consuming and hazardous, and it may cause additional damage. Also, the tension on each vertical suspender rope must be adjusted carefully to distribute the loads of the bridge deck and traffic evenly. Measuring tensile stress in the rope is difficult with conventional technology.

Under a contract from the Advanced Research program, the Southwest Research Institute in San Antonio, TX, developed a nondestructive inspection system based on the principle of magnetostrictive sensing. This system uses simple electrical coils to generate stress waves in steel structures. The waves travel through the stucture and pick up information on the presence and location of defects such as broken strands or corrosion, as well as the tensile stress on the structure. The magnetic fields associated with the stress waves then are picked up by another set of coils.

The Port Authority of New York and New Jersey, and Parsons Transportation Group, used this system on-site at the George Washington Bridge in New York City in the summer of 2000 during the replacement of some of the bridge's suspender ropes. The magnetostrictive sensors monitored the tensioning of the new ropes and also were used successfully to inspect the main suspension cables and the anchorages of these cables. The Port Authority of New York is now planning to apply this technology to other suspension bridges in its system.

Fiber-Optic Sensors Systems

Installation of sensor systems in bridges is increasingly recognized as important for obtaining information on strains, temperature, moisture, and other variables. The information collected from such smart bridges can be used to confirm design calculations, detect damage, and count traffic, among other functions.

An example of the sensor systems developed by the Advanced Research program is the fiber-optic strain gauge based on Bragg gratings. These gratings consist of alternating zones of different indexes of refraction. The spacing of the layers determines a specific wavelength of light that will be reflected. The technology is the same as that used in the broadband fiber-optic telecommunications systems now being installed across the country.

Since the fiber-optic sensor operates with light waves rather than electrons, it has several advantages over conventional electronic strain gauges: ruggedness, absence of drift, and immunity to electromagnetic noise. It permits as many as 100 gauges to be put on a single fiber as thin as a human hair. The installation of the gauges is simplified, the cabling requirement is reduced, and the cost-per-sensor is lowered.

Possible applications may require networks on the order of 1,000 sensors, or 1 kilosensor. Working under an interagency agreement with the Naval Research Laboratory, which has developed many fiber-optic sensors, the Advanced Research program has demonstrated several applications of sensor networks for structural monitoring.

The first application, co-funded with the National Science Foundation (NSF), resulted in the installation of a system of 67 calibrated fiber-optic sensors on an existing steel bridge on Interstate 10 in Las Cruces, NM. This work was carried out by New Mexico State University, with Dr. Rola Idriss as the principal investigator.

"The research has shown the fiber-optic sensors to be a powerful nondestructive evaluation tool," says Idriss. "Whether retrofitted to an existing structure or built into a new smart bridge, they can yield a wealth of information about the structure and the traffic crossing it."

The installation has generated several types of information under random traffic loading, including girder deflections, fundamental vibration frequencies, vehicle speed data, and traffic flow on an hourly basis. To date, the Las Cruces project has achieved notable success in its primary purpose of investigating practical issues in the full-scale application and regular operation of fiber-optic sensors on highway structures. The project has been widely covered in the media and received several awards.

New Mexico State University applied the sensors to the construction of a new concrete bridge in a project co-funded by Advanced Research, NSF, and the New Mexico State Highway and Transportation Department (NMSHTD). The mix design and curing conditions now being used to make high-performance concrete structures may produce unexpectedly high temperatures and stresses during the casting of girders, possibly leading to cracking and major structural failure. Obtaining information on the internal conditions is difficult with conventional temperature or strain gauges because of their fragility.

Forty fiber-optic long-gauge deformation and temperature sensors were embedded in the concrete girders of the Rio Puerco Bridge during casting. These sensors monitored the prestress forces applied to the steel strands in the precast concrete components during and after the steam curing period. One finding was that some design codes considerably overestimate the actual losses. NMSHTD now is planning to use sensors routinely in the construction of concrete bridges in the future. "Building the sensors into new bridges," says Idriss, "enables us to evaluate new high-performance materials and new designs. It also, establishes a baseline for long-term monitoring."

Several companies now offer Bragg fiber-optic sensor systems on a commercial basis. Two States (Hawaii and New Mexico) have received funding from the FHWA Innovative Bridge Research and Construction Program. In addition, several other States are considering installation of these systems on new or existing bridges. Fiber-optic systems also have been chosen as the method for measuring expansion in concrete girders under the lithium treatment evaluation program. All these developments indicate that fiber-optic sensor systems have been transferred successfully from Advanced Research to other FHWA programs.

Advanced Materials Characterization Methods

Although asphalt and portland cement are the most widely used materials in pavements, much remains unknown about the chemical and physical processes that create their respective microstructures and in turn determine the macroscopic properties such as strength and durability. In the absence of this necessary materials science, improving their performance or taking effective measures to prevent deterioration is difficult. A major reason for this knowledge gap has been the lack of suitable techniques to analyze these materials at the required time scales. In order to support the development of the materials science of asphalt and portland cement concrete, the Advanced Research program has explored the application of a variety of innovative methods for materials characterization.

Many of the most promising methods involve nuclear physics and can be done only off-site at dedicated facilities. Some applications have concerned cement hydration, the critical reaction between portland cement powder and water that produces concrete strength, but some have concerned ettringite formation or the alkali-silica reaction, which are deterioration processes.

The most extensive collaboration has been with the National Institute of Standards and Technology's (NIST) Center for Cold Neutron Research on neutron scattering applications. The collaboration has existed for 9 years, almost the entire lifetime of the Advanced Research program itself. Several techniques are involved: Quasi-elastic and inelastic neutron scattering measures chemical bonds, while neutron diffraction and small-angle neutron scattering measure microstructure development. These methods are particularly suited for investigating the hydration of cement because of the strong interactions between neutrons and the nucleus of the hydrogen atoms in water. The techniques are nondestructive, so they can remeasure the same specimen over time to follow the progress of the reaction.

Conducting such measurements in the field on actual concrete structures is not feasible. However, it is possible to use this data to develop mathematical models that predict the rate of hydration as a function of temperature, the water/cement ratio, and so forth. Such mathematical models already are used implicitly or explicitly in a number of concrete standards and tests such as HIPERPAV, the maturity method, and even the 28-compressive strength test.

This neutron scattering research has confirmed that the mathematical models have the general form of the nucleation and growth model that occurs widely in metallurgy and polymer science. This confirmation makes it possible to draw on the vast amount of research in these fields to refine the materials science models of concrete and extend them to new types of high-performance concrete. As a result of the FHWANIST collaboration, researchers have become aware of the applications and have begun to adapt the mathematical models to those applications. The National Science Foundation has awarded several grants to universities that use the models. Industry also has expressed interest in using these methods to evaluate new chemical additives like accelerators and superplasticizers to optimize concrete performance.

Neural Networks

Soon after formation of the Office of Advanced Research, a focused project using neural networks was developed. An initial series of lectures on the subject was followed by an International Conference on Neural Networks in Transportation at The George Washington University. As a continuation to this initial activity, a direct application was funded that demonstrated feasibility of a warning system for drowsy drivers based on neural network methods.

The demonstration used available data on drivers, some of whom had been without sleep for up to 60 hours. The neural networks could detect incipient sleep approximately 3 minutes before a driver actually went to sleep at the wheel. Plans for follow-on work are under development.

Cellular Automata Modeling

The Advanced Research program's interest in this area originated from research noted in the movie animation field called "Boids," now referred to as "emergent behavior." The original animation problem was how to draw flocks of flying birds or herds of running animals efficiently. The same drawing rules devised for the cartoon object seemed directly applicable to drivers in a traffic stream. This form of microscopic modeling for traffic has been used in the highway community previously. However, the Boids algorithm seemed more efficient than those used in traffic models.

The program funded a simple car-following model using the Boids approach. The work provided two unexpected benefits. While testing the simple car-following model, a Turner-Fairbank researcher detected an error in an existing rural two-lane traffic model, and the error has since been corrected.

Also, the literature on Boids contained a reference to modeling people-flow in fire evacuations. This modeling seems immediately applicable to design modeling of toll plazas and modeling of pedestrian/left-turning vehicle interaction.

Multidimensional Data Visualization and Data Mining

Many researchers at Turner-Fairbank have data sets with multiple variables requiring understanding and control of the process that generated the data. As a support to those researchers, the Advanced Research team initiated a project intended to provide access to more powerful data visualization tools with emphasis on depicting the interactions of many variables.

The initial topic under that long-range project focused on a vehicle pollution visualization tool. The out-put displays the pollution from each vehicle in the traffic flow instead of an average over a large network or corridor as determined from sensors. Such a detailed level of information provides a non-compliant region or city with indicators as to where to best impose corrective actions.

Future work in data visualization will explore alternate ways to look at energy flow from finite element simulation output and data extraction from large databases by pattern recognition methods.

Structural Optimization

Most existing structural optimization tools focus on adjustment of member size or material characteristics, usually for linear elastic and static or steady-state conditions. Roadside safety structures are both nonlinear and impulsively loaded, making optimization difficult. This project of the Advanced Research program is focused on optimization of the member size and material parameters, as well as the general global form of the structure.

A first attempt at structural form optimization has been completed at The George Washington University Although limited due to computer run time costs, it is available for exploring small problems. Continued work on structural form optimization is utilizing meshless finite elements. The goal is to develop an efficient method that has a lower-cost factor.

Application of Transform Methods

Initial work in this area focused on detecting roadside hardware from States' photo-log data. The concept was to develop a tool to help manage the highway infrastructure. That initial effort resulted in an algorithm to detect and locate stop signs in video data. This capability is important for maintenance planning and assessment.

Plans are underway to expand this concept to detect and assess more types of infrastructure elements. Work also is ongoing to employ other types of transforms for early detection of cracks or damage in the highway infrastructure and for understanding structural response to earthquakes.

Advanced Research Pays Off

Congress and the research community have recognized the importance of sustaining an Advanced Research program dedicated to applying cutting-edge science and technology to solve critical highway problems. This research may require long-term efforts, so the payoffs may come for future generations. Nevertheless, as shown by these examples, the first decade of Advanced Research has already produced results that are being used in the field. The next decade will bring more.

Main Research Areas by Office

Operations & Safety      Infrastructure

Operations Research      Materials Science
Sensor Applications      Deterioration Science
Articicial Intelligence  Nondestructive Characterization
                          Methods
Analytic Tools           Structural Health Monitoring

Advanced Materials Characterization Facilities Used for Cement Analysis

Facility                  Technique                Application

University of the Ruhr    Heavy Ion Beam           Cement hydration
(Bochum, Germany)

National Institute of     Neutron Scattering       Cement hydration and
Standards and Technology                           micro-structure;
                                                   fly ash reactivity

Los Alamos National       Neutron Diffraction      Ettringite crystal
Laboratory                                         growth

Lawrence Livermore        Positron Annihilation    Alkali-silicate (ASR)
National Laboratory                                gel structure

Lawrence Berkeley         X-ray Transmission       Cement hydration
National Laboratory       Microscope

Argonne National          Synchrotron Diffraction  Ettringite crystal
Laboratory                                         growth

Richard A. Livingston is a senior physical scientist in the Office of Infrastructure R&D at the Turner-Fairbank Highway Research Center (TFHRC) in McLean, VA. His educational background includes a bachelor's degree in history from Dartmouth College (1968), a master's degree in nuclear engineering from Stanford University (1970), and a Ph.D. in geology from the University of Maryland (1990). His professional interests concern the materials science and nondestructive testing of construction materials. During his career he has worked in research positions at the U.S. Atomic Energy Commission, the U.S. Environmental Protection Agency, and the National Institute of Standards and Technology. He also has served as a consultant for the conservation of several architectural monuments, including the Statue of Liberty, the Washington National. Cathedral, Colonial Williamsburg, Westminster Abbey the Taj Mahal, and the Hagia Sophia in Istanbul.

Milton "Pete" Mills is an electrical engineer in the Office of Safety R&D at TFHRC. His educational background includes a bachelor's degree in electrical engineering from North Carolina State University (1963) and a master's degree from Catholic University (1975). At TFHRC since 1968, he has managed the development and evaluation of vehicle sensor systems. He has one patent on a magnetic gradient method for sensing vehicles. His current interests include sensor development and application, application of image processing methods, and development and application of the Super Equation Shell software and numerical error propagation. From 1963 to 1966, he tested and evaluated aircraft antenna systems at the U.S. Naval Air Test Center, Patuxent River, MD. From 1966 to 1968, he designed and patented a number of spacecraft antenna systems at NASA's Goddard Space Flight Center in Greenbelt, MD.

Morton S. Oskard is a structural research engineer in the Office of Safety R&D at TFHRC. His educational background includes a bachelor's degree in civil engineering from the University of Connecticut (1957), a master's degree in engineering mechanics from Northeastern University (1965), a D. Engr. in applied mechanics from Catholic University of America (1980), and postgraduate work in operations research at The George Washington University (1984-89). His current interests are in the areas of structural optimization methods applied to roadside safety structures, multidimensional data visualization, and the application of transforms and neural networks to engineering problems. Prior to his present position, he worked on various aerospace and military structural research projects in the private sector.